Next: Alexandros ...
Up: paciorek-gisspat
Previous: Measurement error
- change of support
- data at different scales: e.g., local exposure information and area-based
health outcome statistics
- try to use your data at their natural scales without aggregating
- interpolation vs. smoothing
- be careful of extrapolation
- cross-validation to check/compare models
- leave-one-out
- n-fold (say 10-fold)
Chris Paciorek
2004-03-18